Progress and Reaction Mechanism of Co-Based Catalysts in the Selective Hydrogenation of α,β-Unsaturated Aldehydes

Abstract

In recent years, Co-based catalysts have attracted considerable attention in research on selective hydrogenation reactions because of their mild activities and favorable selectivities for producing intermediate products, especially in the selective hydrogenation of α,β-unsaturated aldehydes (UAL). However, the low activity of Co-based catalysts for activating hydrogen limits their application in industry, and the diversity of forms and electronic states of Co-based catalysts also leads to the development of complex products and hydrogenation mechanisms at Co active sites. This review provides a comprehensive and systematic overview of recent progress in the selective hydrogenation of UAL over Co-based catalysts, where the preparation methods, hydrogenation properties, and UAL hydrogenation mechanisms of Co-based catalysts are carefully discussed. The influences of nanosize effects, electronic effects, and coordination effects on Co metal and Co oxides are investigated. In addition, the different reaction mechanisms at Co active sites are compared, and their strengths and weaknesses for C=O hydrogenation are further proposed. Finally, the outlook and challenges for the future development of Co-based hydrogenation catalysts are highlighted.

1. Introduction

Compared with traditional homogeneous hydrogenation technology, the selective hydrogenation of α,β-unsaturated aldehydes (ULAs) is still an important process for producing unsaturated alcohols (ULOs) in industry because of its advantages in terms of being environmentally friendly and simple [1,2,3]. In the fine chemical industry, catalytic hydrogenation technology is widely applied in the fabrication of numerous valuable products, such as flavors, drugs, dyes, perfumery, and functional polymers [4,5,6,7]. The hydrogenation pathways of UAL are shown in Figure 1A. The selective hydrogenation of UAL with conjugated double bonds is a complicated reaction because it involves various hydrogenation products, such as unsaturated alcohol (UOL), saturated aldehyde (SAL), and saturated alcohol (SOL), in which UOL is most difficult to obtain because of the disadvantage of C=O hydrogenation regarding both thermodynamics and kinetics. In addition, the byproducts of acetals and ethers may be generated by strong acid sites, leading to complicated products [8,9,10]. Thus, efforts to achieve highly efficient hydrogenation in order to produce UOL still face severe challenges.

Two main pathways for the selective hydrogenation of UAL via heterogeneous catalysis have been reported: H₂ dissociation hydrogenation and catalytic transfer hydrogenation using H₂ and organic alcohols as hydrogen donors, respectively [11]. Traditional H₂ dissociation hydrogenation relies on H₂ pyrolysis to activate H species at the metal center, and the hydrogenation process usually follows the Horiuti–Polanyi mechanism [12,13,14]. Although noble metals, such as Pt, Pd, and Au, have superior activity in UAL hydrogenation [15,16,17,18,19], the drawbacks of high cost and non-renewability limit their development in the future. Thus, researchers are concentrated on studying nonnoble metal catalysts and overcoming the problem of disequilibrium in terms of catalytic and hydrogenation selectivity.

As another hydrogenation pathway of UAL, catalytic transfer hydrogenation is a safe and convenient technology for producing UOL because it circumvents the use of H₂ under high-pressure conditions or the use of other hydride-containing reducing agents [20]. As shown in Figure 1A, hydrogenation and dehydrogenation products are generated simultaneously in the transfer hydrogenation pathway, which is similar to homogeneous hydrogenation via Meerwein–Ponndorf–Verley (MPV) reduction via an isopropanol hydrogen donor [21,22,23]. Organic alcohols work as both solvents and hydrogen donors, resulting in a significant solvent effect on transfer hydrogenation. However, the transfer hydrogenation mechanism and the changes in the reaction interface over catalysts remain controversial and unclear.

In recent years, nonnoble metal catalysts, including Ni, Co, Cu, and Fe, have been widely used in UAL hydrogenation because they are more economical and sustainable than noble catalysts [24,25,26,27,28]. In addition, several design principles for hydrogenation catalysts have been proposed in previous reviews, including turning the electronic structure, the nanosize effect, metal–support interaction, and the steric effect of metal catalysts [29,30,31,32], which are beneficial and practical for most supported-metal catalysts. In addition, metal oxides play an important role in UAL hydrogenation because of the abundant Lewis acid sites originating from M–O structures and the potential synergistic effect of the M⁰ and M^δ+ sites. Moreover, many transition metal oxides and derivative catalysts, such as Co [33,34,35,36,37], Al [38,39], Mg [3,40], and Zr [41,42,43], have been reported for their catalytic activity in transfer hydrogenation. Among them, Co is usually introduced into hydrogenation catalysts as a second metal or assistant for enhancing the C=O hydrogenation properties. When Co is used as the primary active component in selective hydrogenation, it has mild activity and high affinity for C=O in UAL hydrogenation because its special d-band structure is similar to that of noble metals, and Co has seven d-electrons and larger d-bandwidths than other nonnoble metals [44]. Furthermore, Co₃O₄ catalysts are reported to have a selectivity for UOL of more than 90% in UAL transfer hydrogenation [33,34]. As shown in Figure 1B, Co and Co oxide catalysts have satisfactory hydrogenation selectivity for C=O compared with Pt, Pd and Ru noble metal catalysts. In particular, Co oxide catalysts exhibit more than 90% selectivity for UOL in transfer hydrogenation and are regarded as promising alternatives to replace noble metal catalysts. Although numerous studies have been conducted to improve the preparation and modification methods of Co-based catalysts, they still face the problems of poor activity and stability, restricting their large-scale application in industry. With the gradual maturation of advanced characterization and theoretical calculation methods, a deeper understanding of the process from catalyst preparation to the hydrogenation mechanism is essential and meaningful for Co-based catalysts applied in the hydrogenation industry.

Many researchers have comprehensively reviewed the design strategies, reaction properties, and applications of supported metal catalysts in recent years [11,29,30,31,32,45,46], but overall discussions of the preparation, reactivity, and mechanism of Co-based catalysts are incomplete and unclear. Therefore, this review focuses on the strengths, shortcomings, and size scale of the particles in different synthesis methods for Co-based catalysts. The characteristics of UAL-selective hydrogenation on various Co-based catalysts were subsequently discussed and summarized. More attention has been given to the design principles and modification methods of Co metal and Co oxide catalysts. In addition, the main hydrogenation pathways and mechanisms, including H₂ dissociation and catalytic transfer hydrogenation, are highlighted. Finally, ongoing challenges and opportunities for the future development of Co-based hydrogenation catalysts are discussed.

2. Preparation Methods and Size Scales for Co-Based Catalysts

The methods used to prepare Co-based catalysts have crucial effects on the dispersion, particle size and crystal characteristics of Co and Co₃O₄ nanoparticles (NPs). For the hydrogenation activities of metal particles, a theoretical study suggested that the chemical nature of the particle size effect may result in a mutation in hydrogenation reactions when the metal size reaches the cluster scale, but the mechanism of the size effect remains a matter of study [47]. As shown in Figure 2, the nanosize effect may affect the hydrogenation selectivity by changing the adsorption modes of the substrates. With increasing Co NP size, the electronic structure of the Co metal gradually varies from the positively charged (M^δ+) state to the metallic state (M⁰) due to different levels of metal–support interactions. The size of Co NPs can be approximately distinguished at the nanoscale (5~100 nm), cluster scale (0.5~5 nm), and atomic scale [48]. Owing to the difference in structure and the number of aggregated atoms of metal oxides, the size of cobalt oxide clusters has not been clearly defined. Usually, the size and exposed crystal plane of Co NPs can be observed and analyzed via X-ray diffraction (XRD) and transmission electron microscopy (TEM), but metal clusters may not be directly detected and observed via the present analysis methods. Currently, traditional preparation methods and many advanced methods are widely studied and developed, and the strengths, shortcomings and particle scales of different preparation methods are summarized in

Table 1. Various preparation methods for Co-based catalysts.

Synthesis Method	Co Species	Advantages	Disadvantages	Scale
Impregnation method	Co NPs	Operational convenience; facile operation; economy	Need pretreatment; aggregation of particles	Nanoscale
Precipitation method	Co NPs, Co3O4	Economy; simple equipment/process; facile operation	Poor uniformity	Nanoscale
Solvothermal methods	Co3O4	Operational convenience; controllable size; good dispersion	Low production; dependence on equipment;	Nanoscale
Sol–gel method	Co NPs, Co3O4, composite oxides	Mild preparation conditions; controllable structure; anti-deactivation	Long preparation cycle	Cluster scale
Pyrolysis method	Co3O4; Co complexes	Good dispersion; atomic control	Rely on substrate material; long preparation cycle	Atomic or cluster scale

.

2.1. Impregnation Method

As the most facile and widely used method to prepare hydrogenation catalysts in industry, the impregnation method has the advantages of operational convenience and economy [32,49]. For supported catalysts, the substrates are first dispersed in a solution with metal ions, which are adsorbed on the supports by surface tension and solvation effects, and the solvents are removed to obtain the Co catalyst precursor. Typically, Ma et al. prepared a series of Pd/CNT, Co/CNT, and Pd_0.1Co/CNT catalysts via an improved impregnation method [50], in which Pd/CNT and Co/CNT catalysts prepared by the equal-volume impregnation method had particle sizes of approximately 10.9 ± 12.6 nm and 5.3 ± 1.5 nm, respectively. The Pd_0.1Co/CNT catalysts prepared via the successive impregnation method, co-impregnation- method, and deep eutectic solvent impregnation method presented similar particle sizes of approximately 11.2 ± 3.0, 9.4 ± 2.9, and 10.4 ± 3.0 nm, respectively. In addition, the limited structures can easily regulate the size of Co particles via the impregnation method; for example, the pores of MOFs can restrict the growth of Co particles to an average size of 9.0 nm [51]. Modification of the impregnation conditions can control the dispersal of metal particles on the support at the nanoscale. The impregnation method is appropriate for preparing most supported catalysts, but weak metal–support interactions may result in the migration and aggregation of metal NPs.

2.2. Precipitation Method

As one of the widely used production methods in industry, the precipitation method has advantages of economy, simple equipment/processing, and facile operation. In the precipitation method, precipitants, such as ammonia, ammonium, carbonates, and alkalis, are used to immobilize the metal ions on the supports, and the catalysts are finally obtained after the following calcination and reduction procedures [52,53,54,55,56]. The pure Co₃O₄ NPs with a cubic spinel structure prepared via the precipitation method with a NaOH precipitant have a particle size of 3~29 nm, which can be controlled by varying the synthesis conditions [57,58]. Although Co₃O₄ catalysts sintered at high temperatures may have larger particle sizes [59], the introduction of surfactants and thermostable supports efficiently prevents the agglomeration of Co NPs [60,61].

The precipitation method is also the simplest method for preparing multimetal catalysts and composite oxide catalysts because most nonnoble metals can be deposited in basic environments. A familiar case is Co-Ni bimetallic catalysts [56,62,63,64,65], which are widely studied and applied in heterogeneous hydrogenation. Yang et al. prepared a CoNi-LDH hydrotalcite precursor bimetallic catalyst supported on Al₂O₃ by a precipitation method using a mixture of NaCO₃ and NaOH, and the sediments were further calcined and reduced with H₂ to obtain a Co-Ni/Al₂O₃ catalyst [62], in which the average size of the particles was 8.6 nm for Ni₂Co₁/Al₂O₃ and 11.4 nm for Co/Al₂O₃ (Figure 3), demonstrating that the precipitation method is feasible for synthesizing small NPs. A similar particle size of Ni-Co also existed on the NiCo/MgAlO catalyst with a uniform particle size of 14.1 nm, and the study revealed that the particle size further increased with increasing calcination time and significantly affected the hydrogenation properties of FAL [56]. In addition, the Ni-Co alloy may form during calcination and improve the interaction of Ni with Co, preventing the loss and aggregation of metals [63]. Although the precipitation method for preparing Co-based catalysts has obvious advantages in industrial production, the poor uniformity of sediments and particles due to the continuous change in systematic concentration results in regional differences in catalyst properties.

2.3. Solvothermal Method

The solvothermal method is another important synthesis technology for preparing Co-based catalysts; it uses an aqueous solution system with metal ions, complexing agents, precipitants and surfactants and creates internal high-temperature and high-pressure reactors [35,66,67]. The solvothermal method usually involves several processes, including preparing the solution system, dispersing templates or supports, adjusting the pH of the solution, performing a solvothermal reaction, washing and drying the obtained sediments, and calcining. Because the main products of the solvothermal method are cobalt oxides, the reduction of Co catalysts by H₂ or borohydrides before use is necessary to form active metal sites or metal/metal oxide interfaces [68].

Compared with the precipitation method and impregnation method, the solvothermal process can produce more uniform particles of Co oxides. As shown in Figure 4A, the Ru/Co₃O₄ catalyst was synthesized through co-precipitation, followed by the hydrothermal method, where the Co hydroxides were oxidized under mild hydrothermal conditions at 120 °C and formed Co₃O₄ particles [69]. According to the HRTEM and particle size distribution results (Figure 4B–E), the exposed crystal facets of Ru/Co₃O₄ suggest that spinel cubic oxide single-crystalline phases are present in a sample with frequently separated lattice fringes of 0.24, 0.44, 0.23, and 0.28 nm belonging to Co₃O₄ oxide, and the Co₃O₄ particles have an intensive distribution in the range of 10~22 nm, with an average value of 17.5 nm, which is close to the Co particle size calculated via the Scherer equation. To obtain smaller Co₃O₄ NPs, different solvents are used to create additional interactions between Co₃O₄ NPs [70]. For example, Yun et al. prepared a Co@SiO₂ catalyst via a solvothermal process using ethanol as the solvent, followed by a calcination process [71]. The Co@SiO₂ catalyst has an average particle size of approximately 10 nm, suggesting that ethanol may inhibit the growth of cobalt nuclei.

In addition, the morphology of Co₃O₄ can be controlled by alkali additives during the hydrothermal process, helping to generate Co hydroxides or complexes and further resolve Co oxides. Sharad et al. synthesized Co₃O₄ cubes, rods, and sheets via different additives of NaOH, Na₂CO₃, and ammonia [66]. The characterization results revealed that although the crystal structure, surface composition, and reducibility of Co₃O₄ catalysts with different morphologies are highly similar, Co₃O₄ with reactive (110) planes has superior catalytic activity for CO₂ hydrogenation compared with other structures, indicating that the nanostructure and specific exposed facets of Co₃O₄ can improve catalytic activity. In addition, the hydrothermal conditions and solution composition also influence the exposed crystal facet and interface of Co₃O₄, which has been demonstrated as one of the core problems for its catalytic activity [66,68,72].

2.4. Sol–Gel Method

As an advanced method for synthesizing inorganic materials, the sol–gel technique has many advantages in catalyst preparation because of its mild preparation conditions, controllable microstructure of active components, thermal stability, and anti-deactivation ability. The sol–gel method is a wet chemical method used to prepare metal catalysts by forming homogeneous gels, which results in a high dispersion of metal precursors at the cluster level and even at the atomic level, especially for the preparation of mesoporous nanomaterials [73]. The preparation of Co-based catalysts by the sol–gel method can be briefly summarized in three steps: hydrolysis and polymerization of Co active monomers to form sol, evaporation of the solvent and drying gel, and further calcination or reduction to produce a Co catalyst. Furthermore, several key factors for the synthesis of metal catalysts by sol–gel methods need to be considered, including pH, reactant stoichiometry, gelation temperature, solvent, drying and pretreatment conditions [74].

In recent years, the sol–gel technique has been applied to prepare Co-based composite oxides [17,75,76,77,78]. For example, a Co-Al spinel catalyst is prepared via a low-temperature sol–gel method [76], and a schematic illustration of this process is shown in Figure 5A, where the Co and Al nitrates are fully dissolved in ethanol with propylene oxide as a chelating agent and stirred until a Co–Al gel is formed. After aging, the Co-Al gel was dried and calcined under N₂ or air to obtain the Co-Al spinel catalyst. The in situ XRD patterns revealed that the Co-Al gel existed in an amorphous state and that the CoAl₂O₄ phase was generated during high-temperature calcination (Figure 5B,C). Finally, the CoAl₂O₄ particles of SPN-A (calcined in air) had an average size of 8.5 nm. Mitran et al. used different complexing agents, including citric acid, malic acid, oxalic acid, and urea, to prepare CoAl₂O₄ spinels via the sol–gel method and studied their influence on their properties and catalytic activities [79]. The XRD patterns of a series of Co-Al spinels showed similar diffraction peaks for the CoAl₂O₄ and Co₃O₄ phases, and the crystallite size of CoAl₂O₄ calculated via the Scherrer equation was in the range of 10~19 nm. The influence of using different complexing agents may be reflected in the Co³⁺/Co²⁺ ratio and Co²⁺/Al³⁺ ratio, where a good relationship between the Co³⁺/Co²⁺ ratio and acetophenone hydrogenation conversion was observed. The sample prepared with oxalic acid had the largest Co³⁺/Co²⁺ ratio of 1.23 and Co²⁺/Al³⁺ ratio of 4.72, indicating the best hydrogenation activity. Thus, the complexing agents affect the content of high-valence Co sites for Co-Al spinel, which are potential active sites for selective hydrogenation. The sol–gel method has been widely studied and successfully used for preparing uniform composite oxides and derived catalysts. However, gel formation and various procedures for postprocessing lead to longer recycling of catalyst synthesis and high operation costs, restricting industrial production.

2.5. Pyrolysis Method

The pyrolysis method is based on the use of some functional materials or templates to prepare Co catalyst precursors and further form Co catalysts with special structures by thermolysis, which can construct highly dispersed Co or Co₃O₄ NPs and modify their coordination structure to enhance their hydrogenation properties. Thus, the pyrolysis method is usually used in combination with other methods. Typically, the pyrolysis of MOF-based Co precursors results in the formation of a carbon network with anchored Co species, such as ZIF-67 and ZIF-8 [26,35,51,80,81,82,83]. As shown in Figure 6A, the Co-N/C catalyst was first synthesized from the Co-ZIF-67 precursor, which was further pyrolyzed at high temperature to generate Co NPs with an average size of approximately 6–10 nm and many single-atom sites in the carbon network [81]. Furthermore, the use of a surfactant-assisted pyrolysis strategy for Co-based MOFs can further inhibit the agglomeration of Co during the pyrolysis process and afford abundant active CoN_x species and metallic Co clusters (10.8 nm) [83]. Compared with traditional methods, the pyrolysis method results in stronger interactions between the substrate and Co NPs; thus, the prepared Co catalysts have superior reaction properties. In addition, the Co@ZIF-8 (C) catalyst derived from the pyrolysis of the Co@ZIF-8 precursor has a smaller particle size (<2 nm) than Co-ZIF-67 does, indicating that controlling the scale of derived Co particles to reach the cluster or atomic scale by exploring various MOF structures is promising. Moreover, the pyrolysis strategy can also be applied in the template method to realize steric effects and anchoring of metal atoms by physical constraints, and the size of Co metal can be efficiently limited by controlling the structure of the template agents. As shown in Figure 6B, in the synthesis of Co₃O₄ supported on mesoporous carbon, Co²⁺ was first introduced into the polymer gel product by ion exchange, and then Co₃O₄ NPs and the mesoporous carbon network formed simultaneously during the reduction and mild oxidation processes. Ultimately, Co₃O₄ NPs with a mean particle size of approximately 3 nm can be restricted to mesoporous structures [33]. The pyrolysis method for preparing Co-based catalysts has remarkable advantages in terms of steric effects and metal–support interactions, which benefit the dispersion and atomic control of Co metal. However, the catalyst structure after pyrolysis highly relies on the framework and characteristics of the substrate material, and its application in industry may be limited by the long preparation cycle.

3. Hydrogenation Properties of Co-Based Catalysts for UAL

The hydrogenation of UAL is a complicated process over Co-based catalysts because of the diversity of Co species, supports, and interfaces, resulting in complicated intermediates and products. It has been reported that Co⁰ sites and Co single-atom sites have catalytic activity to dissociate H₂ [44,84,85]. Co₃O₄ has been found to catalyze the transfer hydrogenation of UAL with hydrogenous solvents [86], which mainly finishes the transfer of H atoms by forming annular intermediates. In addition, whether other Co-based compounds have hydrogenation activity remains to be studied. The adsorption mode and hydrogenating direction of the substrate are determined by the geometric and electronic properties of the active center [30]. Thus, the catalytic performance of Co-based and Co oxide catalysts in the selective hydrogenation of typical UAL, including cinnamaldehyde (CAL) to cinnamyl alcohol (COL) and furfural (FAL) to furfuryl alcohol (FOL), are summarized in

Table 2. Hydrogenation properties of UAL compared to several advanced Co-based catalysts in recent years.

Entry	Catalyst	Loading (wt %)	Reactant	Product	Reaction Condition				X (%)	SUOL (%)	Ref.
					T (°C)	P (MPa)	t (h)	Solvent
	H2 dissociation hydrogenation
1	Co/TiO2	15.0	CAL	COL	120	1.0	1.0	Methanol	47.4	58.0	[87]
2	CoAl	11.7	CAL	COL	150	-	5.0	Propylene carbonate	51.0	57.0	[88]
3	ε-CoNP/GOX	4.5	CAL	COL	120	2.0	4.0	Dioxane	45.1	100.0	[89]
4	Co-C-500	56.0	CAL	COL	90	2.0	4.0	Ethanol + H2O	85.3	51.5	[90]
5	Co@NPC	69.8	CAL	COL	80	0.5	5.0	H2O	85.5	79.1	[91]
6	Co-NC-450	80.0	FAL	FOL	110	1.5	2.0	H2O	92.0	99.0	[92]
7	Co/NC-700	55.0	FAL	FOL	120	2.0	0.5	Methanol	83.0	100.0	[93]
8	Co/CoNx/C	48.9	CAL	COL	30	2.0	13.0	methylbenzene	94.0	93.0	[26]
9	Co-N-C@F1270.3	29.1	FAL	FOL	140	1.0	2.0	2-POL	97.2	92.5	[83]
10	Co@L2N@b-TiO2–N-7	3.13	CAL	COL	100	2.0	6.0	2-POL	59.0	97.0	[94]
11	Co@BN/BN-600	8.5	CAL	COL	120	0.4	11.0	Ethanol	83.3	81.3	[95]
12	Co2Al-H2-400	NA	CAL	COL	100	2.0	2.0	H2O	97.0	73.6	[96]
13	Co/ZrLa0.2Ox	8.8	FAL	FOL	80	2.0	2.0	H2O	95.0	92.0	[97]
14	CuCo/MgO-0.04	5.0	FAL	FOL	100	2.0	0.7	2-POL	82.4	92.7	[98]
15	CuCo/NC	39.3	FAL	FOL	100	3.0	6.0	Ethanol	35.6	94.3	[82]
16	CoZnB/Nb2CTx	9.5	CAL	COL	100	3.5	2.0	2-POL	91.4	72.6	[99]
17	Co-Fe-1%Zn-B	60.7	CAL	COL	130	2.0	1.0	Ethanol	97.7	100.0	[100]
18	Co-Fe/ACBT	8.3	CAL	COL	120	2.0	5.0	Methanol	90.5	86.8	[101]
19	Fe0.5Co@NC	46.8	CAL	COL	80	2.0	5.0	Water	95.1	91.7	[102]
	Catalytic transfer hydrogenation
20	Co3O4@NC	3.5	CAL	COL	180	1.0	12.0	2-POL	81.1	94.1	[35]
21	Co3O4/MC	15.0	FAL	FOL	160	1.0	6.0	2-POL	100.0	98.0	[33]
22	Co3O4/NCNF	22.0	CAL	COL	160	-	5.0	2-POL	100.0	95.0	[34]
23	CoOx@Co-500	15.6	CAL	COL	120	-	8.0	2-POL	97.5	90.1	[37]
24	ZIF-67@SiO2-CPTEOS	17.4	CAL	COL	180	1.0	12.0	2-POL	84.6	95.0	[103]
25	Co3O4-Al2O3	45.8	FAL	FOL	150	-	6.0	2-POL	76.3	97.0	[104]
26	CoCrOx (1:2)-CN	8.6	CAL	COL	120	1.0	12.0	2-POL	97.5	95.3	[105]
27	Co1.5Mg1.5Al1-LDO-500	15.5	FAL	FOL	160	-	6.0	2-POL	99.7	95.9	[106]

to compare the inherent properties, nanoscale effects, and interfacial effects of Co-based catalysts.

3.1. Metal Catalysts

3.1.1. Monometallic Co-Based Catalysts

Monometallic Co catalysts are simple and efficient catalysts for producing UOCs because of their natural hydrogenation selectivity for C=O. To investigate the inherent hydrogenation activity of Co, a series of monometallic Co catalysts supported on Al₂O₃, SiO₂, and TiO₂ were prepared via the conventional impregnation method for CAL hydrogenation [87], where the Co/TiO₂ catalyst showed the highest CAL conversion of 47.4% and selectivity of 58% for COL, suggesting that the Co NPs preferentially hydrogenated C=O, followed by C=C hydrogenation. Even for the Co-CoO@SiO₂-500 catalyst with a core–shell structure, the selectivity of COL ultimately reaches approximately 65% under steric effects [67]. The above monometallic Co catalysts prepared via a simple method usually generate typical hexagonal close-packed (hcp) Co, which mainly exposes (1 1 $\overline{2}$ 0)-type facets and has higher hydrogenation activity than face-centered cubic (fcc) Co and epsilon (ε) Co. Notably, ε-Co NPs have been successfully prepared, which have an average particle size of 6.5 nm and exhibit superior selectivity to COL (100%), but the ε phase of Co cannot stably exist under the reaction conditions [89].

It is well known that reducing the size of metal particles can increase the specific surface area and the exposed metal sites of catalysts and thus improve their catalytic performance [30,107]. However, it seems that the smaller size of Co NPs is not always beneficial for C=O hydrogenation. For example, a CoAl spinel catalyst derived from layered double hydroxide (LDH) precursors has smaller Co NPs than 3 nm after H₂ reduction [88], which results in moderate selectivity for UOL (~60%) because the adsorption of both C=O and C=C may be allowed simultaneously on small particles of Co, although the adsorption of the C=O bond is predominant. For another hydrotalcite-derived Co₂Al-400 catalyst [96], Co NPs with a mean size of 7.0 nm showed a higher conversion of 97.0% and selectivity of 73.6% for COL in H₂O, suggesting that Co NPs have a significant size effect on their hydrogenation activity. For larger particles over the Co/NC catalyst, Co NPs have an average size of 12.8 nm, which results in better performance in terms of both conversion (83%) and selectivity (>99%) for producing UOL [93], suggesting that smaller Co particles are not necessarily better for hydrogenation. The same nanosize effect also occurs for the adsorption and conversion of other substrates on Co NPs [47,76]. Thus, according to the above discussion, the size of Co NPs should have an appropriate range of approximately 6~20 nm close to the cluster or subnanometer scale for hydrogenation.

To enhance the hydrogenation properties of monometallic Co catalysts, an effective strategy is to introduce functional supports to adjust the size and electronic properties of Co NPs at the same time. Owing to the natural electronic structure of Co metal, its ability to dissociate H₂ can be improved by the use of electron-rich supports to increase the d-electron intensity of Co NPs. For example, Tian et al. prepared a Co NP catalyst supported on a carbon material derived from the MOF precursor Co-BTC [90]. Compared with the other monometallic Co catalysts, the Co-C-500 catalyst resulted in greater conversion of CAL (85.3%) at a low reaction temperature of 90 °C, which was attributed to the uniform dispersion of Co NPs on the carbon support, with an average diameter of 11.3 nm. Furthermore, the doped elements are introduced into the carbon material to modify the electronic structure of the Co NPs. For example, nitrogen is introduced into MOF-derived Co/CoN_x/C catalysts to enhance C=O hydrogenation over Co-N_x active sites [26]. As shown in Figure 7A–D, the Co NPs were uniformly anchored on the derived carbon network with a mean diameter of 6~7 nm, and the Co/CoN_x/C-600 catalyst showed improved selectivity for COL (>90%). The Raman and XPS spectra of the catalyst revealed possible N-bonds over Co/CoN_x/C (Figure 7E), indicating that the Co-N_x sites may work as active centers or promote electron transfer. The electronic effect of doped N has also been reported for other derived carbon materials [91,92,93] and boron nitride materials [95,108], demonstrating that it is a widely practicable strategy for the design of Co-based catalysts. As mentioned above, monometallic Co catalysts are widely used in C=O-selective hydrogenation, and the selectivity for UOL can reach 90% when some functional materials are used as supports. In addition, Co NPs have a size effect, which is reflected by the remarkable decrease in activity and C=O selectivity in UAL hydrogenation when their size is less than 6 nm.

3.1.2. Bimetallic Co-Based Catalysts

Co is usually introduced as an assistant to noble metals, such as Pt, Pd, and Ni, to improve their C=O hydrogenation properties, and the promoting role of Co has been sufficiently discussed in previous reviews [29,30,46]. In this section, the roles of Co and second metals in the structural effect and electronic effect are concentrated. On the basis of the above discussion, monometallic Co catalysts have moderate catalytic activity and preferentially attack C=O, followed by C=C hydrogenation. Therefore, second metals are introduced to further enhance their catalytic properties, which mainly play a role in two aspects: modifying the properties of Co metal and forming synergistic effects with Co via additional functional sites.

In general, Co combined with some nonnoble metals, such as Cu [82,98], Fe [101,102], and Zn [99], may have synergistic effects. For example, a CuCo/MgO bimetallic catalyst synthesized by simple impregnation and calcination methods has different electronic effects on Co and Cu [98], where the electron-rich Cu sites play a role in H₂ dissociation, and the d-band center of the CuCo/MgO catalyst moves upward from −2.4756 eV to −2.4350 eV (Figure 8A,B), resulting in a more pronounced adsorption capacity of the Cu sites for H₂ and intermediate, ultimately resulting in an increasing conversion of 82.4% and selectivity of 92.7% compared with those of the monometallic catalyst. In addition, Xu et al. demonstrated that the improved catalytic performance of the CuCo system was due to the high dispersion of the Cu-related species on the Co NPs, confinement effect of the N-doped carbon supports and synergistic effect [82]. The DFT results suggested that the energy of H₂ dissociation was only 0.003 eV for the CoCu/NC catalyst, and the barrier to the hydrogenation of C=O by active H atoms also decreased (Figure 8C). Importantly, the rate-determining step of UAL hydrogenation over a Co-based bimetallic catalyst may involve the activation of reactants and a progressive hydrogenation process.

The large d-bandwidth of Co and its role in C=O hydrogenation have been proven by many studies. However, Co sites lack the ability to inhibit the adsorption of C=C, resulting in complicated reaction products. In our previous work, the effect of a dilute alloy on a CoZnB/Nb₂CT_x catalyst was studied, where Zn was introduced to modify the coordination and electronic states of CoB NPs [99]. As shown in Figure 9, Zn worked as a dispersing agent and restricted the growth of CoB NPs, and the enhanced activity of CoZnB/Nb₂CT_x at low temperatures (from 15.2% to 91.4%) was attributed to the smaller particle size (reduced from ~40 nm to ~15 nm) and electron transfer effect, which promoted the adsorption and activation of H₂ and C=O at the Co and B sites, respectively. This size effect of metal atoms via the distributed arrangement of multimetal atoms also occurred for Co-Fe [101,102] and Co-Fe-Zn systems [100], which not only weakened the concentrated adsorption of nonpolar C=C on adjacent Co-Co sites but also inhibited side reactions involving solvent effects. For Co-based catalysts, the reductive etherification and aldolization of aldehydes and alcohols can be catalyzed by cobalt and Lewis acid cooperative sites [8,10,109]; thus, decreasing the coordination number and bond length of Co-Co with a second metal efficiently reduces the number of byproducts.

3.2. Co-Based Oxide Catalysts

Co oxide components widely exist in all Co-based catalysts, and their role in catalytic reactions is easily overlooked. It has been reported that the synergistic effect of Co⁰ and Co^δ+ sites over the Co⁰/CoO_x interface can remarkably improve the process of H₂ dissociation and activate hydrogen transfer to substrates [110]. In addition, Co₃O₄ NPs are highly selective catalysts for transfer hydrogenation via the use of organic alcohols as hydrogen sources [33,86]; therefore, Co₃O₄ NPs are safer than traditional H₂ hydrogenation methods under high pressure. Compared with H₂ dissociation hydrogenation, transfer hydrogenation of UAL on Co-based catalysts shows significant differences in two aspects. On the one hand, catalytic activity is determined by the protonation and proton transfer ability of the catalyst for organic alcohols. On the other hand, the solvent effect has a more marked effect on the hydrogenation selectivity. In this section, the catalytic performances of Co-based oxide catalysts are briefly reviewed.

3.2.1. Co Oxide Catalysts

Compared with Co NPs, pure Co₃O₄ NPs have natural properties for transfer hydrogenation and show greater selectivity for C=O. Wang et al. prepared Co₃O₄ NPs (~3 nm) supported on mesoporous carbon (Co₃O₄/MC), which have 95% selectivity toward UOL at full conversion, higher than that of most Co metal catalysts [33]. Notably, a size effect of Co₃O₄ NPs has been reported for CO₂ [111] and acetone [47] hydrogenation, but its influence on the hydrogenation selectivity of UAL is still poorly understood. For FAL and 5-hydroxymethylfurfural hydrogenation, Co₃O₄-6 nm, Co₃O₄-17 nm, and Co₃O₄-nanocasting catalysts are introduced for comparison with Co₃O₄/MC. As shown in Figure 10A,B, the Co₃O₄/MC catalyst with an average Co₃O₄ NP size of 3 nm is more active than the other Co₃O₄ catalysts. In addition, Co₃O₄-6 nm exhibited the best reaction rate (2.29 mmol/(g_cata·h)) at 120 °C compared with Co₃O₄-17 nm (0.26 mmol/(g_cata·h)) and Co₃O₄-nanocasting (0.32 mmol/(g_cata·h)) samples without support [33], suggesting that smaller Co₃O₄ NPs may increase the hydrogenation rate.

Furthermore, Co₃O₄ NPs are more resistant to agglomeration and do not suffer from oxidative deactivation, resulting in better reaction stability than Co NPs under reaction conditions. Although the Co₃O₄ NPs showed superior C=O selectivity in UAL hydrogenation, their poor activity led to strict reaction conditions and a low turnover frequency. Thus, the main factors and modification methods used to promote the intrinsic and total activities of Co₃O₄ catalysts have been studied. For Co-based catalysts supported on carbon materials, N doping is a widely used strategy to improve the interaction between catalysts and substrates [34,91,92,93,112]. For example, Cui et al. successfully synthesized MOF-derived Co₃O₄@NC catalysts for the transfer hydrogenation of CAL [35]. As shown in Figure 10C–J, the Co₃O₄ NPs were uniformly dispersed on the carbon networks derived from the MOFs, and the average particle size distribution of Co₃O₄@NC was 7.5 nm, whereas it was 10 nm for Co₃O₄@C, suggesting that the introduction of N in the MOFs reduced the aggregation of Co₃O₄ NPs. In terms of their hydrogenation properties, the conversion of CAL increased 41.2% after N doping with 95.9% selectivity to COL, which was attributed not only to the change in Co₃O₄ NP size but also to the intervention of functional N sites in the hydrogenation process. In general, doped N sites or functional groups containing N act as basic sites to adsorb active H species or intermediates on solid catalysts [35,113]. Moreover, a specific Co-N_x can modify the electronic structure of Co sites [26,114], ultimately resulting in the promotion of Co-based catalysts.

In addition, the composition and valence state of Co oxides affect their hydrogenation performance. Gong et al. synthesized a simple shell-core structured CoO_x@Co catalyst for UAL hydrogenation to UOL with a selectivity over 90% [37]. As shown in Figure 11A–C, a clear interface of Co⁰/CoO can be observed in the TEM images, with a concentrated distribution of Co in the core. The XPS spectra further verified the reduction of Co oxides and the generation of the Co⁰ core (Figure 11D,E). Moreover, a synergistic effect of Co⁰–Co^δ+ sites over the CoO_x@Co catalyst was revealed by DFT calculations. As shown in Figure 11F, CAL had an adsorption energy of −1912 kJ/mol on the CoO (111) surface, suggesting that the low activity of Co oxides in transfer hydrogenation may be ascribed to extremely strong adsorption, resulting in a decreased turnover frequency and possible hydrogenation of C=C. The internal Co⁰ sites of the CoO_x@Co catalyst can facilitate weak adsorption toward C=O, suggesting that an optimal CoO_x composition is required to simultaneously achieve efficient conversion and high selectivity. This influence of the chemical composition of Co oxides on selective hydrogenation properties lays the foundation for the design of Co-based composite oxide catalysts.

3.2.2. Co-Based Composite Oxide Catalysts

Similarly to Co-based bimetallic catalysts, the drawback of low activity for Co oxide catalysts is overcome by the formation of composite oxide catalysts, including mixed metal oxides and multimetal oxides. As shown in Table 2 entries 24–27, the required reaction conditions for Co-based composite oxide catalysts are milder, while the transformation rate and selectivity for C=O are also superior to those of pure Co oxide catalysts. Mixed or supported metal oxide catalysts are easily obtained via simple impregnation methods and precipitation methods. For example, a hydrotalcite-derived Co₃O₄-Al₂O₃ catalyst was synthesized via the co-precipitation method [104], which showed good activity for FAL transfer hydrogenation and 97% selectivity for C=O at low temperatures. The introduction of Al₂O₃ may provide crucial basic sites for transfer hydrogenation via the MPV reaction. However, the formation of side products derived from the acetalization and etherification pathways is also observed at various temperatures because of the presence of several strong acid or basic sites [8,9,10].

Another strategy to construct Co-based composite oxide catalysts is to dope other metals into the spinel structure of Co₃O₄. For example, CoMgAl composite oxide (LDO) catalysts are derived from Co-based ternary layered hydroxides for transfer hydrogenation of FAL [106]. As shown in Figure 12A,B, the characteristic diffraction peaks of the LDH precursor basically disappeared after calcination, new diffraction peaks belonging to Co₃O₄ spinel appeared, and the intensity gradually increased, suggesting that Mg and Al did not exist as independent oxides. The surface acidity and basicity of Co_1.5Mg_1.5Al₁-LDO-500 significantly increased compared with those of the Mg₃Al₁-LDO-500 catalyst (Figure 12C,D), in which the basic site (O²⁻ site) and Lewis acid sites (Co^δ+) promoted the deprotonation of isopropanol and the activation adsorption of FAL via the MPV route, resulting in an improved hydrogenation activity of 99% conversion and 95% selectivity to FOL. In addition, Mao et al. successfully prepared a CoCrO_x spinel catalyst supported on N-doped carbon via the sol–gel method and revealed the influence of interfacial electron migration and N-doped carbon on the acid–base synergy of the CoCrO_x catalyst [105], in which CoCrO_x(1:2)-CN showed a 95.3% yield of COL and an activation energy of 69.2 kJ/mol for CAL transfer hydrogenation lower than those of other samples (Figure 13A), demonstrating that the Co–Cr spinel phase is more active for transfer hydrogenation. Furthermore, the effects of the Co/Cr ratio and doped N on the acidity and basicity were investigated. As shown in Figure 3B,C, CO₂-TPD and NH₃-TPD profiles revealed that strong acid and basic sites near 597.1 °C and 582.7 °C, respectively, were generated when the Co/Cr ratio increased to 1:1, suggesting the interaction of Co and Cr. After the introduction of N into carbon, both the surface acidity and basicity significantly increased (Figure 13D), indicating that N sites not only provided surface basic sites for the catalyst but also promoted the exposed acid sites from CoCrO_x spinel. Co oxides and Co-based composite oxides seem to be promising catalysts for transfer hydrogenation because of their safe reaction conditions without high-pressure hydrogen and superior selectivity to UOL. However, the effects of the Co oxide composition, interface, and weak interactions of the substrate need more in-depth investigation.

4. Reaction Mechanisms of UAL over Co Active Sites

On the basis of differences in hydrogen donors, the hydrogenation of UAL to produce UOL can be conducted via two pathways: H₂ dissociation hydrogenation and catalytic transfer hydrogenation. As shown in Figure 14, H₂ dissociation hydrogenation usually follows the Horiuti–Polanyi mechanism over a metal center, where H₂ and UAL are adsorbed and activated on adjacent metal sites simultaneously, and the selectivities of products are determined by the adsorption model of conjugated double bonds. The transfer hydrogenation pathway mainly followed the MPV mechanism, in which the C=O of UAL and the hydroxy of organic alcohol can form a ring intermediate and further finish the hydrogenation of C=O by Lewis acid sites. In both hydrogenation pathways, solvents have a crucial effect on the reaction process. On the one hand, solvents play a role in promoting mass transfer and increasing the contact surface area of reactants in direct H₂ hydrogenation [115,116], where alcohols do not participate in hydrogenation. On the other hand, transfer hydrogenation using solvents as hydrogen resources can occur by altering the catalyst and reaction conditions, where the hydroxyl hydrogen of alcohol is activated and transferred to C=O or C=C and generates aldehyde or ketone products. In addition, side reactions between alcohols and aldehydes, such as aldolization [9,10] and reductive etherification [117], may occur during both hydrogenation pathways. Thus, the reaction process, solvent effect, and several potential hydrogenation mechanisms for UAL hydrogenation over Co-based catalysts are reviewed and discussed in this section.

4.1. Hydrogen Dissociation Hydrogenation

Thermocatalytic H₂ dissociation hydrogenation is still one of the most typical and widely applied technologies for modern industry and pharmaceutical applications [118,119] and relies on H₂ pyrolysis to activate H species on active metal centers, metal-base sites, and frustrated Lewis pair (FLP) sites [12,120,121]. As shown in Figure 15, for Co-based catalysts, H₂ can be activated at adjacent Co–M (M = Co or second metal) sites and Co–X (X = base site) sites by homolytic dissociation or heterolytic dissociation. Moreover, the adsorption and desorption behavior of UAL at the active center markedly affects the hydrogenation selectivity and catalytic activity of catalysts [30]. In this section, the characteristics and characteristics of H₂ dissociation hydrogenation of UAL over different Co active sites are summarized.

The H₂ dissociation efficiency usually determines the total catalytic rate of catalysts in hydrogenation. For the Co NPs, the DFT calculations revealed that the dissociation barriers of H₂ on Co(311), Co(111), and Co(110) are 13.0, 2.7, and 14.5 kJ/mol, respectively, which are all lower than their desorption energies, indicating that H₂ is likely activated at the Co active center and that Co(111) is more active for H₂ pyrolysis to activate hydrogen [122]. Furthermore, the influences of the particle size and crystal structure of Co NPs are reviewed. First, the different H₂ dissociation energies that rely on the size of the fcc Co NPs are shown in Figure 16A, in which the H₂ activation efficiency of the Co NPs in the particle size range of 3.5~40 nm according to H–D exchange analyses demonstrated that the more facile HD formation at lower temperatures is, the larger the Co particles become [47], indicating that large Co NPs benefit H₂ dissociation. The influence of the Co NP crystal structure can subsequently be explained by their H₂ desorption ability. As shown in Figure 16B, hcp Co exhibited an obvious desorption peak of H₂ near 100 °C, which was not observed for fcc Co. The amount of H₂ released from hcp Co (21.0 μmol/g) was up to two times greater than that released from fcc Co (9.7 μmol/g) [123], resulting in better H₂ activation efficiency for hcp Co. The activation of H₂ by Co centers may be related to the bond length and structure of the exposed Co–Co sites, indicating differences in the hydrogenation properties of Co NPs with various particle sizes and crystal structures.

On the basis of the discussion above, H₂ can be dissociated by dual active sites in principle. However, it is exceptional for single-atom catalysts. For example, Cao and coworkers prepared a single-atom Co catalyst with a g-C₃N₄ template and adjusted its coordination structure via P-doping [113]. As shown in Figure 16C, the single-atom Co₁-N-C and Co₁-N/P-C catalysts had H₂ desorption temperatures of 535 K and 697 K, respectively, which are higher than those of the other Co NP catalysts, indicating that the stronger interaction and activation of H₂ are related to the coordination environment of the single-atom Co sites. Furthermore, the dissociation processes and energies of H₂ over Co₁-N-C and Co₁-N/P-C catalysts are calculated via DFT simulation, where one of the H atoms attaches to the Co center while the other H atom transfers to the neighboring N atom or P atom, and the dissociation energy barriers are only 2.06 eV and 0.71 eV for the Co₁-N-C and Co₁-N/P-C catalysts, respectively (Figure 16D). It seems that H₂ activation over Co₁–N₄ or Co₁–N₂P₂ sites followed the Lewis acid-base principle; the greater the charge difference between acid–base sites was, the easier the dissociation of H₂ was. Although single-atom Co catalysts exhibit superior activity in hydrogenation, their application in UAL hydrogenation has not been reported.

The adsorption structures of CAL on Co sites are shown in Figure 17A,B, where a vertical configuration can be observed on the Co (111) and CoO (200) planes with adsorption energies of −0.58 eV and −5.64 eV, respectively [91], indicating that the Co oxides have relatively strong adsorption on CAL, resulting in possible overhydrogenation. The moderate adsorption of C=O on Co (111) allowed the COL product to desorb. However, unmodified Co NPs also adsorb C=C via concentrated Co–Co sites, leading to a horizontal adsorption configuration of UAL and the imperfect selectivity of the target UOL product [124]. The doped N can efficiently improve the interaction between Co NPs and C=O. As shown in Figure 17C, differential charge density analysis revealed that the N-doped carbon support provided more electrons to transfer from the N to the Co NPs, resulting in an electron-rich state and a negative charge on the Co NPs [26]. Furthermore, the Co/CoN_X/C-600 catalyst showed a lower energy barrier for the first hydrogenation of C=O (0.07 eV) than for C=C hydrogenation (0.39 eV) and C=O hydrogenation over the Co/C catalyst (0.43 eV) (Figure 17D,E), ultimately resulting in an enhanced low-temperature activity of 100% conversion and 78% selectivity for C=O.

In addition, H₂ and UAL hydrogenation are simultaneously activated at the Co active center, which increases the difficulty of the highly selective hydrogenation of C=O. Therefore, the M (M = second metal) and X (base site) sites are introduced to form dual metal sites with Co, and this change in structure and electronic structure directly influences the H₂ dissociation activity and adsorption mode of substrates over the Co center. As shown in Figure 8C, the dual Co–Cu site had a lower energy barrier for H₂ dissociation and the migration of H to C=O, whereas a stronger vertical configuration of C=O (1680 cm⁻¹) and two negligible characteristic peaks of the furan ring (1580 cm⁻¹ and 1450 cm⁻¹) were observed in the in situ FTIR spectra of the Cu-Co catalyst than in those of the single Cu catalyst (Figure 18), indicating that the introduction of Co into Cu inhibited the production of byproducts. The enhanced performance of CuCo catalysts is due mainly to electron transfer between Co and Cu [82,98], which results in the formation of new bimetallic sites with different electronegativities favoring the heterolytic dissociation of hydrogen [12]. The same electronic effect is appropriate for other electropositive metals, such as In and Ga, which lose their electrons and serve as active sites for activating polarized C=O groups [124].

The hydrogenation dissociation mechanisms at Co–X sites are quite different from those at Co–M sites because of the lack of adjacent metal–metal bonds. When the neighboring atom of a metal center is replaced with a nonmetal atom or ligand, H₂ may be activated via a dihydrogen complex pathway to form active H species. First, the electron transfers from the filled σ orbital of H₂ to the empty d_σ orbital of the metal center to form a two-electron, three-center orbital bonding over all three atoms [125,126]. Then, a dihydride complex (H^δ−–M–H^δ+) is generated by the dissociation of H₂ because the d_π electron of transition metals is donated back to the σ^* antibonding orbital of hydrogen [125,127], ultimately resulting in a proton–hydride pair (M–H^δ−/X–H^δ+) consisting of a metal center (Lewis acid) and a Lewis base with heterolytic dissociation of H₂ [128]. This heterolytic process is similar to the hydrogenation mechanism in FLP systems [120,129,130,131].

For Co NPs, the UAL is easily adsorbed on the electropositive site via polarized C=O, whereas the active H can be dissociated on dual metal sites and directly attack C=O to form an intermediate, leading to a decrease in the energy barrier. In addition, the Co–X sites on the surface or interface of catalysts can promote the spatial separation of the adsorption site of UAL from the H₂ dissociation site, which allows good tunability of the reaction selectivity. However, the reduced density of exposed active sites may be the cost of high hydrogenation selectivity, leading to poor low-temperature activity for Co-based bimetallic catalysts and metal-based catalysts compared with Co-NP catalysts (Table 2).

4.2. Catalytic Transfer Hydrogenation

The catalytic transfer hydrogenation of UAL over metal catalysts mainly abides by the acid–base catalysis principle for organic alcohols and unsaturated compounds, which results in superior properties for C=O-selective hydrogenation [21,42]. Because organic alcohols are used as hydrogen donors, the key step for the transfer hydrogenation mechanism is the activation of hydroxy groups and the generation of activated intermediates from organic alcohols and UAL.

Unlike H₂ dissociation hydrogenation, catalytic transfer hydrogenation mainly occurs over Co oxide catalysts, where the Co^δ+ sites work as active centers [34,37]. Cui et al. carried out DFT calculations to study the differences in potential active sites over the Co₃O₄@NC catalyst for transfer hydrogenation, confirming the energy profiles of the CAL hydrogenation pathways on Co₃O₄ (111), N-Co₃O₄ (111), Co₃O₄ (111)-O_v, Co…N FLP sites, and Co…O FLP sites (Figure 19A,B) [35]. The Co₃O₄ (111) surface and unliganded Co (111)-O_v site had a higher energy barrier for CAL transfer hydrogenation but markedly decreased at the unliganded Co…O FLP site. Furthermore, the Co…N FLP site exhibited a lower activation energy than the Co…O FLP site did, indicating that stronger basic sites can promote the dissociation of O–H and the formation of intermediates [34]. Moreover, the catalytic transfer hydrogenation over Co–O and Co–N sites seems to follow the Meerwein–Ponndorf–Verley (MPV) pathway [20,21], as shown in Figure 19C. The C=O group of CAL and the OH group of the organic alcohol may be adsorbed on the Co^δ+ site simultaneously and form a cyclic six-membered intermediate. Furthermore, Cui et al. revealed the transformation process of similar cyclic six-membered intermediates on Co–N centers over a Co-based MOF catalyst [103]. As shown in Figure 19E, the H_a in isopropanol first transferred to the adjacent N from the Co–N bond, and then the Co–N was broken. Subsequently, another H_b in isopropanol is transferred to C_a in CAL, and isopropanol forms acetone. Finally, a new isopropanol was adsorbed on the Co center and provided its H_a to O_a in CAL to generate a target product of COL. Another pathway involves the adsorption of C=O and OH groups on adjacent M–O–M sites [105,106]. For example, Cheng et al. proposed a potential hydrogenation pathway for CoMgAl-LDO mixed oxide catalysts [106]. As shown in Figure 19D, the H atom of the organic alcohol is adsorbed on the Lewis basic site (O²⁻), thus weakening the O–H bond, and the Lewis acid site (Co^δ+) can directly activate C=O. Proton transfer from alcohol to O²⁻ subsequently promotes the formation of a six-membered intermediate between Co, isopropanol, and FAL. This six-membered ring mechanism of UAL transfer hydrogenation over Co-oxide catalysts revealed the direct effect of C=O on the Lewis acid site, which is responsible for the high hydrogenation selectivity to C=O.

Compared with H₂ dissociation hydrogenation, catalytic transfer hydrogenation over Co-based catalysts showed superior hydrogenation selectivity to C=O due to the strong interaction of C=O and Lewis acid sites. However, the Co oxides showed poor activity for catalytic transfer hydrogenation at low temperatures. Tuning the coordination structure of Co centers to further adjust the surface acid–base sites of Co-based catalysts may be an effective strategy to increase their activity for catalytic transfer hydrogenation. In summary, H₂ dissociation hydrogenation and catalytic transfer hydrogenation of UAL over Co-based catalysts have significant differences as well as some similarities. On the one hand, H₂ and organic alcohol as hydrogen donors are activated to produce active H species on the Co–Co sites and Co–X sites, respectively, with different energy barriers and reaction conditions. On the other hand, C=O is similarly adsorbed and activated on Co centers or Lewis acid sites, determining the direction of hydrogenation. In addition, the synergistic effect of Co⁰ and Co–X may further enhance the hydrogenation performance [37,110,114], but new aspects of hydrogenation mechanisms require further understanding.

4.3. Solvent Effects and Side Reactions

In this section, the influences of solvents are briefly discussed to further understand the hydrogenation process and potential side reactions. In general, UAL hydrogenation reactions commonly use low-carbon organic alcohols as solvents, including methanol, ethanol, and isopropanol, because of their better capacity for dispersing substrates and proton transfer than other solvents [132]. The hydrogenation properties are involved in the polarity of the solvent molecules. First, protic solvents with strong polarity have high H2 solubility and a lower barrier for exchanging hydrogen [14,115]. For example, a series of protic solvents sorted according to their polarity, i.e., water > methanol > ethanol > isopropanol, were investigated for their influence on polarity over Co-Fe catalysts supported on acidic bentonite [101], in which the methanol group resulted in higher conversion and selectivity for C=O, demonstrating that the high polarity of the solvent is beneficial for H₂ hydrogenation at Co⁰ active sites. Moreover, when a solvent molecule with a long carbon structure is adsorbed on the catalyst surface, it may have a steric effect and reduce the interaction between UAL and the active sites, leading to poor hydrogenation activity of the catalyst [133]. With respect to transfer hydrogenation, the solvent effect showed the opposite trend [34,105,106]. For example, Wang et al. studied the influence of hydrogen donors on the transfer hydrogenation of CAL, where the reaction activity decreased with decreasing polarity of C₂–C₅ alcohols over an Al₂O₃ catalyst, which was attributed to weaker alcohol–CMA and alcohol–catalyst interactions in low-polarity solvents [38]. For the CoMgAl mixed oxide catalysts, as shown in Figure 20A, with increasing carbon chain length, both the conversion and C=O selectivity increased in FAL transfer hydrogenation [106], indicating that a specific polarity may favor the catalytic transfer hydrogenation reaction for C=O.

Furthermore, the influence of solvent polarity on the selectivity of hydrogenation was investigated. It has been reported that nonpolar solvents prefer to activate nonpolar C=C, and polar solvents benefit from attacking C=O [134]. To further reveal the relationship between solvent and selectivity over Co-based catalysts, Gao and coworkers studied the performance of Co₂Al-H₂ catalysts with different solvents [96]. As shown in Figure 20B, the C=O selectivity of the catalyst obviously increased with increasing solvent polarity, which revealed that strongly polar protic solvents promoted proton migration and intermolecular hydrogen bonding, resulting in increased UOL productivity. However, this regularity of protic solvents was not distinct over the Co@Pd–Co core–shell catalyst, where the hydrogenation selectivity was similar when H₂O, methanol, ethanol, and isopropanol were used [50]. This may be because the adsorption of solvent is weak on the catalyst surface; thus, the hydrogenation properties are mainly determined by the interaction of reactants and catalysts [83].

Moreover, water is a particular solvent for UAL hydrogenation. On the one hand, water can be easily activated and transfer protons to produce UOL due to its extremely strong polarity compared with that of organic alcohols; meanwhile, some side reactions involving organic alcohols can be avoided [10]. On the other hand, UAL is difficult to dissolve in water, resulting in a lower mass transfer rate of H₂ and a smaller contact area between the reactant and the catalyst. Thus, using a mixed solvent with an appropriate proportion of water can enhance the selective hydrogenation of UAL [109,133]. According to the above results, although the nature of the solvent clearly influences the hydrogenation performance of Co-based catalysts, the magnitude of the solvent effect on hydrogenation is strongly dependent on the catalyst identity, hydrogenation pathways, and nature of the solvent employed, which may result in different interactions between the adsorbed solvent and unsaturated double bonds in the reactant and transition states on the metal surfaces.

Another important impact of the solvent effect is side reactions involving organic alcohols and UAL, including aldolization and reductive etherification. The unstable aldehyde functional group can be catalyzed by acid sites to form acetals via organic alcohols [9], which protects the aldehyde group from hydrogenation and reduces the hydrogenation rate. Therefore, unsaturated acetals and saturated acetals from UAL and SAL may be produced on surface acid catalysts during hydrogenation [10,135,136,137]. For organic alcohols, methanol and ethanol are preferable for the formation of acetals because polarized hydroxyl groups are easily broken and dehydrated. The pure Co⁰ or Co–O–Co active sites cannot catalyze side reactions, but the synergy of Co and Lewis acid sites will occur [8,77]. On the mixed Co-based catalysts, aldolization easily occurs on metal Lewis acid sites from surface acid supports and the metal–support interface because the simultaneous adsorption of C=O and hydroxy groups on acid sites is necessary for producing acetals, making it easier to compete with transfer hydrogenation.

In recent years, side reactions during UAL hydrogenation involving organic alcohol solvents have been reported, but the mechanism and its relationship with the structure and active sites of catalysts remain ambiguous. To avoid the production of byproducts, several solutions have been proposed. First, because the side reactions of aldolization and reductive etherification are reversible, appropriate reaction conditions, such as increasing the concentration of substrates and H₂ pressure, are selected to enhance the hydrogenation reaction. Second, a mixed solvent of alcohol–water was prepared, where the appropriate amount of water can reduce the contact between UAL and alcohols and promote the hydrolysis of acetals. Third, the use of solvents with long carbon structures results in steric effects, which can weaken the adsorption of solvents on the surface of catalysts.

5. Conclusions and Perspectives

In summary, Co-based catalysts play a pivotal role in the C=O-selective hydrogenation of UAL to produce UOL due to the special electronic structure of Co metal. Thus, the preparation methods, hydrogenation properties, and UAL hydrogenation mechanisms of Co-based catalysts are carefully summarized in this review. The discussions suggest the crucial role of particle size, bimetallic synergistic effect, and coordination structure for Co NPs and Co oxide catalysts. First, the strengths and shortcomings of different preparation processes and the corresponding particle scale and nanosize effects on the hydrogenation properties of Co NPs are considered, where the optimal range of Co particle sizes for obtaining high conversion and selectivity simultaneously is approximately 6~20 nm. A value either above or below this range may result in the loss of catalytic activity or C=O selectivity for Co-based catalysts. Second, the coordination structure and electronic properties of Co can be regulated by introducing a second metal or ligand, further enhancing the catalytic activity and interaction of C=O over Co⁰ and Co^δ+ sites. Third, the UAL hydrogenation mechanism of H₂ dissociation hydrogenation and catalytic transfer hydrogenation over Co-based catalysts are summarized, which suggests the key role of Co active sites in the adsorption and transformation of hydrogen donors and UAL.

Moreover, in light of these two hydrogenation mechanisms, the key factors and following strategies are proposed to improve Co-based catalysts. As shown in Figure 21, for the Co NP catalyst, the first mechanism regulates the particle size of the metal in an appropriate range of approximately 6~20 nm because of the nanosize effect. The second mechanism modifies the electronic structure of the Co metal to achieve an electronic effect by introducing a second metal or ligand. Finally, the steric structure around the Co center should be controlled to restrict the adsorption of C=C and the benzene skeleton. For the Co oxide catalyst, catalytic transfer hydrogenation also had a nanosize effect on both the hydrogenation activity and selectivity. The solvent effect is remarkable for catalytic transfer hydrogenation due to the strong interaction of the catalyst and solvents, which is related to the polarity of organic alcohols. Finally, tuning the surface acid and base sites is essential because the Lewis acid–base effect simultaneously affects the transfer hydrogenation activity and the side reactions of aldolization and reductive etherification.

Co-based catalysts are promising for replacing noble metal catalysts because of their high hydrogenation selectivity, but many practical problems and challenges need to be solved. First, the poor activity of Co-based catalysts at low temperatures results in higher production costs and longer production cycles. Although many advanced Co-based catalysts, such as single-atom catalysts and intermetallic components, have been rapidly developed in recent years, Co-based catalysts lag behind noble metal catalysts. Thus, improving the intrinsic catalytic activity of Co centers is crucial in the future. Furthermore, the study of the hydrogenation mechanism over Co-based catalysts is insufficient, especially for the catalytic transfer hydrogenation of UAL, which constrains researchers’ ability to design high-performance catalysts. The complexity of the hydrogenation process involving solvents and the role of the Co/CoO_x interface require further investigation via advanced in situ and operando characterization techniques. Finally, the stability of Co active sites should be carefully considered when designing catalysts because of the agglomeration, loss, and poisoning of metal and metal oxide NPs. The construction of functional supports for forming strong metal–support interactions and protective structures may be an efficient strategy to improve the stability of Co active sites.